Research On Dynamic Discrimination Of Traffic State Based On Floating Car Data

Traffic State Estimation(TSE)is critical to transportation planning,operation and infrastructure design.In order to manage the traffic flow and possible traffic congestion,the essential step is to measure,estimate,and predict the states of traffic flow in the traffic network,such states could be,the traveling speed,the number of vehicles in the network,whether or not the road is congested,and whether or not there is an traffic incident.In this new era of “computer times” and the development of information technology,massive volumes of sensing data from a variety of source(such as cell phones,GPS,probe vehicles,and inductive loops,etc.)enable Intelligent Transportation System(ITS)in an efficient,timely and accurate manner.Traffic State Estimation is to estimate the characteristics of traffic flows in road traffic.Based on the data pre-processing of GPS floating car,this paper studies urban traffic roads and studies the dynamic Traffic State Estimation method.This research develops a Bayesian-based theoretical framework,along with statistical inference algorithms,to capture the complex flow patterns in the urban traffic road network,to enable both estimation and perdition of traffic states and to estimate the traffic state dynamically.To achieve these goals,a Hierarchical Bayesian probabilistic model is proposed to capture spatial-temporal traffic states.The model discretizes time and space,and uses the Hierarchical Bayesian network to graphically describe the temporal and spatial changes of traffic flow in a two-dimensional coordinate system.Through the macroscopic fundamental diagrams method and the intrinsic correlation of traffic flow parameters,the key variables in t he relational expression are pointed out.The propagation of traffic states are encapsulated through the Link Queue Model and equilibrated fundamental diagrams.The Expectation Maximization Extended Kalman Filter(EM-EKF)algorithm is used to infer the unknown traffic state under the Hierarchical Bayesian model.It is expected that the M-step in the Expectation Maximization algorithm aims to estimate the optimal model parameters,and the E-step aims to estimate unobserved traffic states by utilizing Gaussian approximate inference.Finally,to test the proposed methodology,we applied the EM-EKF algorithm on VISSIM software and in the real world,the arterial network area in Hongta District,Yuxi City.The small simulation experiment proves that EM-EKF could have the potential to make traffic state estimation in an urban arterial network.The Yuxi results show that estimated traffic data approaches the true data as time proceeds without considering the weather conditions and traffic accidents.